This work aims to develop methods to identify functional sites in protein structures and to characterize protein function on a genomic scale. The approach is predicated on the Evolutionary Trace method (ET) to locate functional sites in structures. Preliminary studies enabled us to automate the basic steps towards a complete, automated functional annotation pipeline, namely, functional site analysis with ET; extraction from ET analysis of SD-templates that describe composition and conformation of key residues involved in binding or catalytic function; the search in other structures for geometric matches to these 3D-templates; and the analysis of which of those matched are most biologically relevant. We now seek to increase the sensitivity and specificity of the annotation pipeline by optimizing the definition of 3-D templates, by adding new template features to better judge whether molecular mimicry underlies functional similarity; and by developing novel strategies that use multiple templates to identify function. The result will reveal which regions of proteins are most biologically relevant, and hence logical targets for protein engineering and drug design, and it will extend to three dimensions a functional annotation strategy traditionally based on one- dimensional pattern matching in protein sequences. In so doing, this work addresses a fundamental NIH roadmap problem in "post-genomic biology": linking massive and exponentially growing amounts of raw sequence and structure data to the molecular basis of biological function. [unreadable] [unreadable] [unreadable]